Fullstack AI Platform Engineer

Position Summary

We are seeking an Intermediate Fullstack AI Platform Engineer to help build production AI agents and agent-backed workflows on AWS.

This is a hands‑on engineering role for someone who has built real software around LLMs or agents, not just experimented with prompts or demos. You will work across Python backend services, React/Type Script interfaces, Amazon Bedrock, Agent Core‑style runtime patterns, tool calling, structured outputs, retrieval workflows, evaluation, observability, and integrations with internal systems.

The role sits between AI application engineering and AI platform engineering. You will build scoped agent features while also contributing to reusable patterns that help future agents become easier to build, test, deploy, observe, and maintain.

This is not a research role, data science role, prompt‑writing role, or chatbot‑only role. We are looking for engineers who can ship production software, debug failures, work with APIs and backend systems, and explain what they personally built.

• Build and maintain production AI agents and agent‑backed workflows using Python, AWS, and modern agent frameworks.
• Implement agent logic for planning, tool selection, tool execution, structured responses, multi‑step workflows, and error handling.
• Preference for candidates who have built AI, LLM, RAG, or agent‑backed workflows used by real users, internal teams, customers, or production‑like environments.
• Build integrations between agents and internal APIs, databases, enterprise systems, retrieval sources, and external tools.
• Implement structured output patterns using JSON Schema, Pydantic, validation, retries, and response normalization.
• Work with Amazon Bedrock or comparable managed LLM services for model invocation, inference configuration, prompt handling, and response processing.
• Support Agent Core‑style runtime patterns, including session handling, runtime invocation, memory‑aware workflows, execution metadata, and agent observability.
• Build RAG workflows using embeddings, vector search, document chunks, metadata filters, retrieval tuning, and source attribution.
• Contribute to MCP‑style tool/server integrations and multi‑agent handoff patterns where applicable.
• Build Python backend services for agent execution, API integration, job processing, session state, response persistence, and debugging.
• Build React/Type Script screens for testing agents, reviewing outputs, managing configuration, viewing evaluations, and monitoring execution status.
• Write automated tests for agent behavior, tool calls, structured outputs, retrieval workflows, backend APIs, and frontend flows.
• Support evaluation workflows using test datasets, expected outputs, regression checks, model‑based scoring, and human review.
• Troubleshoot real production issues involving tool failures, malformed outputs, retrieval quality, hallucinations, latency, cost, observability gaps, and integration errors.
• Work with senior engineers to implement features within established AWS, CI/CD, security, and observability patterns.

• 3–5 years of experience as a fullstack, backend, AI application, platform‑adjacent, or infrastructure‑mindful software engineer.
• 1–3 years of Python backend engineering experience.
• Hands‑on experience building or integrating AI, LLM, RAG, or agent‑backed applications.
• Ability to walk through at least one real AI/LLM/agent project in detail, including architecture, users, tools/integrations, failure modes, and what you personally owned.
• Experience with at least one agentic framework or orchestration approach such as Strands, Lang Graph, Lang Chain, Semantic Kernel, Auto Gen, CrewAI, or comparable tools.
• Understanding of agentic application patterns: tool calling, structured outputs, planning, multi‑turn workflows, session state, memory, retrieval, and evaluation.
• Experience defining or consuming structured outputs using JSON, JSON Schema, Pydantic, OpenAPI, or similar validation approaches.
• Experience integrating applications with REST APIs, internal services, external tools, databases, or enterprise systems.
• Practical exposure to RAG, embeddings, vector databases, semantic search, document…